LLR ENHANCEMENT MODEL FOR A TURBO EQUALIZER

The turbo equalizer (TEQ) is a baseband signal processor that combines the tasks of channel equalization and channel decoding in a closed loop by using turbo principles. The basic objective of a TEQ is to take care of interference in a communication system arising due to a host of reasons and correct the random errors caused by additive white Gaussian noise (AWGN) simultaneously. All the TEQs reported in literature can be broadly categorized into three types; the trellis based equalizers, the apriori aided minimum mean square error (MMSE) filters and the soft interference canceller (SIC) type of filters. Pivotal to the operation of all the TEQs is the flow of the extrinsic information between the equalizer and the channel decoder. The extrinsic information is the information about a given bit available by the use of knowledge about other bits in a transmitted data stream. Most often, the extrinsic information is modeled as a Gaussian random variable. However, a generic mathematical model capturing the operations of a TEQ seems to be missing in literature. It is with this motivation that we propose a suitable mathematical model for a TEQ under converging conditions. This is the log likelihood ratio (LLR) enhancement model that describes the evolution of the extrinsic information as observed at the equalizer output. The LLR becomes a monotonically increasing function of the signal to noise ratio (SNR) from the no apriori condition to the perfect apriori condition about the transmitted bits. KEYWORDS: